"Geoffrey S. Mendelson" wrote:

DaveM wrote:

Sometimes the mfrs make a line of transistors having very similar
characteristics, differing in only a few specific elements, such as Vce,
Ic(max), etc.

Sometimes they took the same transistor and tested it. Depending upon
the results, it became one part or the other.

Back when the IBM PC first came out you could buy memory that cycled
at 200, 150, 125 or 100 ns. There was only one production line.

Each chip was tested at 100ns (10mHz), if it passed it was maked -100
and sent off to be packaged. If it passed at 125ns (8mHz) it was marked
-125, and so on. If it did not pass at the lowest speed the chips were
sold at it was tossed.

Now the chips that failed would be sold on eBay as "untested". :-)



Really? Then explain the 4164 chips of various brands marked 300
that tested to 150 or better on my RAM tester. I repaired a LOT of
Commodore 64 computers, and bad or slow memory was one of the biggest
failures. I salvaged thousands of used 4164 and 41256 chips from Unisys
mainframe memory boards. All that were still good (over 99%) tested at
least one level faster than they were marked. They told me that they
bought millions of RAM chips, speed tested them, and then sold the
slower culls to the IBM clone makers and on the spot market.